Variability in 14C contents of soil organic matter at …eprints.gla.ac.uk/136600/1/136600.pdf3428 T. S. van der Voort et al.: Variability in 14C contents of soil organic matter at

Biogeosciences, 13, 3427–3439, 2016www.biogeosciences.net/13/3427/2016/doi:10.5194/bg-13-3427-2016© Author(s) 2016. CC Attribution 3.0 License.

Variability in 14C contents of soil organic matter at the plot andregional scale across climatic and geologic gradientsTessa Sophia van der Voort1, Frank Hagedorn2, Cameron McIntyre1,3, Claudia Zell1, Lorenz Walthert2,Patrick Schleppi2, Xiaojuan Feng1,4, and Timothy Ian Eglinton1

1Institute of Geology, ETH Zürich, Sonneggstrasse 5, 8092 Zürich, Switzerland2Swiss Federal Research Institute WSL, Forest soils and biogeochemistry, Zürcherstrasse 111,8903 Birmensdorf, Switzerland3Department of Physics, Laboratory of Ion Beam Physics, ETH Zurich, Otto-Stern-Weg 5, 8093 Zurich, Switzerland4State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences,Beijing 100093, China

Correspondence to: Tessa Sophia van der Voort ([email protected])

Received: 20 December 2015 – Published in Biogeosciences Discuss.: 18 January 2016Revised: 12 May 2016 – Accepted: 13 May 2016 – Published: 14 June 2016

Abstract. Soil organic matter (SOM) forms the largest ter-restrial pool of carbon outside of sedimentary rocks. Radio-carbon is a powerful tool for assessing soil organic matter dy-namics. However, due to the nature of the measurement, ex-tensive 14C studies of soil systems remain relatively rare. Inparticular, information on the extent of spatial and temporalvariability in 14C contents of soils is limited, yet this informa-tion is crucial for establishing the range of baseline propertiesand for detecting potential modifications to the SOM pool.This study describes a comprehensive approach to exploreheterogeneity in bulk SOM 14C in Swiss forest soils that en-compass diverse landscapes and climates. We examine spa-tial variability in soil organic carbon (SOC) 14C, SOC con-tent and C : N ratios over both regional climatic and geologicgradients, on the watershed- and plot-scale and within soilprofiles. Results reveal (1) a relatively uniform radiocarbonsignal across climatic and geologic gradients in Swiss foresttopsoils (0–5 cm,114C= 130± 28.6, n= 12 sites), (2) simi-lar radiocarbon trends with soil depth despite dissimilar envi-ronmental conditions, and (3) micro-topography dependent,plot-scale variability that is similar in magnitude to regional-scale variability (e.g., Gleysol, 0–5 cm, 114C 126± 35.2,n= 8 adjacent plots of 10× 10 m). Statistical analyses haveadditionally shown that 114C signature in the topsoil is notsignificantly correlated to climatic parameters (precipitation,elevation, primary production) except mean annual temper-ature at 0–5 cm. These observations have important conse-

quences for SOM carbon stability modelling assumptions, aswell as for the understanding of controls on past and currentsoil carbon dynamics.

1 Introduction

Soil organic matter (SOM) constitutes the largest terrestrialreservoir of carbon outside of that held in sedimentary rocks(Batjes, 1996). With on-going rapid changes in land use andclimate, carbon stability within this reservoir may be subjectto change (Davidson and Janssens, 2006; Melillo et al., 2002;Schimel et al., 2001; Trumbore and Czimczik, 2008). Theprocesses that lead to both stabilisation or destabilisation ofSOM, however, remain poorly understood (von Lützow et al.,2008; Schmidt et al., 2011; Trumbore and Czimczik, 2008).The existence of numerous biological and physicochemicalprocesses that operate simultaneously confound our under-standing of, and ability to accurately model SOM dynamics(Lutzow et al., 2006).

Radiocarbon measurements of SOM in combination withcompositional data are increasingly used to assess SOMturnover rates and residence time, often coupled with mod-elling (Davidson and Janssens, 2006; Schrumpf et al., 2013;Sierra et al., 2012; Tipping et al., 2010, 2011; Torn et al.,2005; Trumbore, 2009). The marked 14C enrichment asso-ciated with the radiocarbon “bomb-spike” (Stuiver and Po-

Published by Copernicus Publications on behalf of the European Geosciences Union.

3428 T. S. van der Voort et al.: Variability in 14C contents of soil organic matter at the plot

lach, 1977) aids in the elucidation of processes taking placeon decadal scales, while natural 14C decay can shed light onprocesses that occur on timescales of centuries to thousandsof years. Although a powerful tracer, it is presently not wellestablished how the amplitude of 14C variations associatedwith carbon turnover and storage are masked by spatial het-erogeneity. Soils are intrinsically highly heterogeneous overa range of spatial scales (Goovaerts, 1998), and effects ofclimatic, geological and topographical gradients as well assmall-scale variability at the plot or pedon scale are oftenoverlooked or generalized. While spatial variability in soilshas been explored in the context of SOM compositional pa-rameters (Conen et al., 2004; Goovaerts, 1998; Grüneberg etal., 2010), systematic studies of its influence on 14C charac-teristics of SOM are lacking. Moreover, despite the high car-bon stocks in subsoils, deeper soil carbon characteristics havebeen much less intensively compared to topsoils, and there-fore remain poorly understood, particularly in the context of14C variability (Fontaine et al., 2007; Rumpel and Kögel-Knabner, 2011).

In order to examine the temporal and spatial scales of 14Cvariability in soils, we have investigated SOM 14C signaturesin depth profiles across diverse climatic, and geologic gra-dients, as well as over different spatial scales. These find-ings provide a framework for anticipating and accountingfor spatial variability that are of importance in developingsampling, measurement and carbon-cycle modelling strate-gies. The broader goal is to provide the foundation for un-derstanding, and ultimately predicting, regional-scale SOM(in)stability based on radiocarbon characteristics.

2 Materials and methods

2.1 Study area and plot selection

The study area, which covers the entirety of Switzerland, islocated between 46–47◦ N and 6–10◦ E with an accompa-nying elevation gradient between roughly 480 and 1900 mabove sea level (a.s.l.). Within this geographic and altitu-dinal range, mean air temperature (MAT) and mean annualprecipitation (MAP) varies from 1.3 to 9.8 ◦C and ∼ 600 to2100 mm yr−1, respectively (Etzold et al., 2014; Walthert etal., 2003). Samples investigated in this study derive fromthe Long-term Forest Ecosystem Monitoring (LWF) networkmonitored and maintained by the Swiss Federal ResearchInstitute of Forest, Snow and Landscape Research (WSL)(http://www.wsl.ch/lwf). The LWF sites are all located inmature forests and samples were collected during the 1990sand have been stored in containers of hard plastic (Innes,1995). They encompass marked climatic and geologic gra-dients, yielding an array of different soil types (Fig. 1). Sup-plement regarding these sites (e.g. clay content and profiledepth) can be found in Supplemental Table S1. The LWF-network was chosen because it features (i) archived samples

Figure 1. Locations of the sampled sites within Switzerlandindicated by white points. Numbers refer to the locations1Alpthal, 2Beatenberg, 3Lausanne, 4Nationalpark, 5Othmarsingen,6Bettlachstock, 7Isone, 8Jussy, 9Lens, 10Novaggio, 11Schaenis,12Visp and 13Vordemwald (Innes, 1995).

(ii) environmental and ecological monitoring data and (iii)long-term monitoring and sampling potential. An overviewof the most relevant lithological, geomorphic and climatevariables of the sampled sites can be found in Table 1. Soilsare classified in accordance with the World Reference Base(WRB, 2006).

2.2 Sampling methodology

Samples were taken in the course of the 1990s in a 43 by43 m (∼ 1600 m2) plot, that was divided in quadrants of 20meters (∼ 400 m2) and sub-quadrants of 10 m (∼ 100 m2),each with 1 m of walking space in between (Walthert et al.,2002; Fig. 2a). For depth increments down to 40 cm soil ofan area 0.5 by 0.5 m was sampled (0.25 m2), for samples> 40 cm corers were used (∼ 2× 10−3 m2). Because differ-ent soil types were included, depth increments rather thangenetic horizons were analysed. In order to consider the plot-scale spatial variability, three mixing schemes were devised.

1. Eight discrete samples per depth were taken fromthe sub-quadrants according to a chessboard-pattern(Fig. 2a).

2. Sixteen samples per depth were taken from sub-quadrants, and combined and mixed for each quadrantof the plot (1–4, 5–8, etc), yielding in total four com-posite samples (Fig. 2b).

3. Sixteen cores are taken from sub-quadrants, all samplesare combined and mixed to yield a single compositesample (Fig. 2a).

4. The destructive nature of digging soil profiles down tothe unweathered parent material only allowed for singleprofiles to be taken proximally to the intensively moni-tored plot.

Biogeosciences, 13, 3427–3439, 2016 www.biogeosciences.net/13/3427/2016/

http://www.wsl.ch/lwf

T. S. van der Voort et al.: Variability in 14C contents of soil organic matter at the plot 3429

Table 1. Overview of WSL LWF sampling locations and main climate variables. Mean average air temperature (MAT) and mean averageprecipitation (MAP) and Net Primary Production (NPP) measured between 1980 and 2010. Numbers in first column correspond to sitenumeration of Fig. 1. Soil classification in accordance with WRB (2006), altitude in m a.s.l., soil and slope as a percentage of 90 ◦ (Etzold etal., 2014; Walthert et al., 2003).

Soil type Geologya Locationa Slopea Altitudea

MATb MAPb NPPb

( ◦C) (mm yr−1) (g C m−2 yr−1)

1 Gleysol Flysch Alpthal 11.5 1149–1170 5.3 2126 4452 Podzol Sandstone Beatenberg 16.5 1490–1532 4.7 1163 3023 Cambisol Moraine with calcareous and shale fragments Lausanne 3.5 800–814 7.6 1134 8244 Calcisol Alluvial fan of calcareous character Nationalpark 5.5 1890–1907 1.3 864 1115 Regosol Moraine of calcareous character Othmarsingen 13.5 467–500 9.2 1024 8456 Cambisol Calcareous sediments Bettlachstock 33 1101–1196 6.4 1273 5307 Cambisol Moraine Isone 29 1181–1259 7.2 2126 4908 Stagnosol Moraine of calcareous character Jussy 1.5 496–506 9.4 961 5049 Calcisol Colluvial deposits Lens 37.5 1033–1093 8.1 969 28910 Umbrisol Moraine Novaggio 34 902–997 9.5 1545 58011 Cambisol Calc supported conglomerate Schaenis 30 693–773 8.5 1691 77012 Regosol Calc supported colluvial deposit Visp 40 657–733 9.8 595 8613 Stagnosol Moraine Vordemwald 7 473–487 8.1 1123 774

a Walthert et al. (2003), b Etzold et al. (2014).

1 2

43

5 6

7 8

43 m

43 m

A

10 m

1 2

3 4

5 6

7 8

9 10

11 12

13

15

14

16

43 m

43 m

B

10 m

Figure 2. Soil sampling scheme according to Walthert et al. (2002)at the Long-Term Forest Ecosystem (LWF) Research Program,(a) eight individual samples taken in a chessboard pattern, (b) 16samples that are combined to mixed samples in each plot quadrant,as indicated by the red boxes (1–4, 5–8, etc).

Soil samples were dried in an oven at 35–40 ◦C, sievedthrough 2 mm, and stored in a climate controlled storage fa-cility at the WSL. In order to determine the density of the fineearth, volumetric samples were sampled in single profiles.These samples were taken using steel cylinders of 1000 cm3

volume (for layers with a thickness of at least 10 cm) or458 cm3 volume (for thin layers with less than 10 cm thick-ness). Volumetric samples were dried at 105 ◦C for 48 h min-imum until the resulting mass remained constant. The den-sity of the fine earth was determined based on oven-driedvolumetric soil samples and sieving the samples in a waterbath to quantify the weight of stones > 2 mm. The volume ofstones was calculated by assuming a density of 2.65 g cm−3

for stones (Walthert et al., 2002).Variability within and between two similarly sized plots

was investigated in a subalpine spruce forest in the Alp-tal Valley (central Switzerland), separated by approximately

500 m. The plots are located within the same watershed andhosted on the same bedrock and are primarily distinguishedby their degree of tree coverage.

The two plots concerned are the following: (1) WSL LWFplot (∼ 1600 m2, sampled 1998), covered mainly by Nor-way Spruce trees. Sampling was done by depth-incrementin a 10× 10 m grid (Fig. 2b); (2) WSL NITREX con-trol plot, (∼ 1500 m2, sampled 2002) with a relatively opentree canopy with well-developed herbaceous vegetation andshrubs (Schleppi et al., 1998). This site was sampled by hori-zon (litter, topsoil and Gleyic horizon) on an 8× 8 m grid,yielding 24 soil cores. These cores were grouped accordingto their morphology.

Interpolation based on documented soil cores was usedto convert the NITREX control plot horizon-logged cores todepth-logged cores.

2.3 Elemental, 13C and 14C analysis

Vapour acidification (12 M HCl) of soils in fumigated silvercapsules was applied to remove inorganic carbon in the sam-ples (Komada et al., 2008; Walthert et al., 2010). Carbonate-poor samples were acidified for 24 h at room temperature, forcarbonate-rich samples this treatment was extended to 72 hat a temperature of 60 ◦C. Both acidification treatments weretested on the same sample and no significant effect on theradiocarbon values was found. To ensure the removal of allcarbonaceous material, δ13C values of the soil samples weremeasured. All glassware and silver capsules were combustedat 450 ◦C (6 h) prior to use.

Graphitization of the soil organic matter was performedusing an EA-AGE (elemental analyser-automated graphitiza-tion equipment) system (Wacker et al., 2009). The radiocar-bon content of the graphite targets was measured on a MI-

www.biogeosciences.net/13/3427/2016/ Biogeosciences, 13, 3427–3439, 2016


CADAS (MIniturised radioCArbon DAting System) at theLaboratory of Ion Beam Physics, ETH Zürich (Wacker et al.,2010). Samples were calibrated against oxalic acid II (NISTSRM 4990C) and an in-house anthracite coal as well as an in-house soil standard (Aptal soil 0–5 cm). Intra-sample (Alptalsoil 0–5 cm) repeatability is high with a value and standarddeviation of Fraction Modern (Fm)= 1.1263± 0.0003696(n= 12).

An Elemental Analyser-Isotope Ratio Mass Spectrome-ter system (EA-IRMS, Elementar, vario MICRO cube – Iso-prime, Vison) was used to measure the absolute concentra-tions of carbon and nitrogen, as well as stable carbon isotopiccomposition (δ13C). Calibration standards used for the C andN concentrations, and C : N ratio were peptone (Sigma) andatropine (Säntis) and only atropine (Säntis). The δ13C val-ues were used to assert that all inorganic carbon has beenremoved by the procedure, as well as provide a confirmationof the signature of the plant input.

2.4 Turnover model and statistical analysis

For the turnover time or mean residence time (MRT) esti-mates the system was assumed to be in steady state, with notime lag between atmospheric carbon fixation by plants andconversion to soil carbon. A time-dependent model (Eq. 1)(Herold et al., 2014; Torn et al., 2009) was used in combi-nation with atmospheric data of Levin et al. (2010). Here,k refers to the decomposition rate constant and λ to the ra-dioactive decay constant of 14C in yr−1 (1/8267). Rsample isinferred from 114C as shown in Eq. (2) (Herold et al., 2014;Solly et al., 2013).

Rsample =k×Ratm,t

+ (1− k− λ)×Rsample ( t-1 ) (1)

Rsample =114Csample

1000+ 1 (2)

MRT was calculated for the samples for which the signal wasabove the atmospheric 114C signal of the sampling year, in-dicating a predominant input of post-bomb derived carbon.The constant k is found by matching the modelled and mea-sured values of Eq. (1), and MRT is obtained by invertingk (Torn et al., 2009). For the 0–5 cm topsoil layer two MRTwere frequently possible, in which case it was assumed thetrue MRT value of the deeper layer is the one that exceedsthe MRT value of the accompanying litter layer, as carbonturnover rates decrease with increasing soil depth.

Statistical analyses were performed in R, version 3.1.2.Spatial variability of 14C was quantified by calculating thecoefficient of variation (CV) of the Fm 14C values. The Fmvalues rather than the 114C values were used as CV tendstoward infinity if the average value of the variable concernednears to zero, which in this case would cause unwanted bias

(Stuiver et al., 1997). Drivers of CV were analysed by Pear-son correlation (rcorr function; hmisc package) using ancil-lary climate and geographic data taken from the WSL LWFnetwork (Walthert et al., 2003). Ancillary data entailed MAT,MAP, net primary production (NPP), pH and mineralogicalcomposition such as clay, sand and silt content. The Spear-man correlation coefficient (rcorr function; hmisc package)was used to test relationships between Fm and climatic andcompositional parameters for data from separate depth inter-vals. The effects of climate and soil properties on radiocar-bon (114C) was also tested by fitting mixed-effect modelsby maximum likelihood (lme function; nlme package; (Pin-heiro and Bates, 2000). Mixed-effect model analyses providean additional benefit because they eliminate variance intro-duced by random variables, whereas this is lacking in e.g.the Spearman correlation. For the mixed-effect model, thecomplete data set was used, i.e. all depths, soil cores and allindividual depths were included. The variables site and corewere always taken to be random effects following the sam-pling design at the LWF sites. Environmental drivers (MAT,MAP, NPP, pH, clay content) were taken as fixed effects andtested individually. The compositional parameters (e.g. clay,pH) are depth interval-specific. For the analysis concerningall depths, 114C was tested against depth and the additionalparameters. For set depth intervals, 114C was related onlyto one parameter per analysis. In all final models, normalityand homoscedasticity of the residuals were verified visuallywith diagnostic plots; the dependent variables were all logtransformed in some cases to meet assumptions of normalityand variance homogeneity. Semivariograms (variog function;geoR package) were also used to assess the patterns of vari-ability with distance.

3 Results

Spatial variability was systematically investigated from theregional (100–300 km) to the plot (10–100 m) scale.

3.1 Regional scale (100-300 km) variability

At the LWF sites across Switzerland (Table 1), the standarddeviation in 114C values of the organic (LF) (n= 12), 0–5 cm (n= 12) and 5–10 cm (n= 11) layers exceeds the in-strumental error by up to tenfold with values of 36.4, 28.6and 34.9 ‰, respectively (Table 2). Furthermore, the coeffi-cient of variation in SOC content increases with depth whilst114C variability remains almost constant (Fig. 3). The MRTrelative error remains constant with depth (Table 2). For theformer and latter analyses, a Podzol site (Beatenberg) wasexcluded because its unique thick organic layer distorted thetopsoil age. The Spearman coefficient identified few signif-icant correlations between 114C in samples and climaticvariables. Exceptions include a negative correlation between114C vs. MAP and vs. SOC in the 10–20 cm depth interval



0

20

40

600 50 100 150 200 250

Δ14C

0–5 cm 5–10 cm

Organic layer

SOC

(%)

Figure 3. Relationship of 114C vs. soil organic carbon content(SOC) on the regional scale within Switzerland (n= 12 sites), witheach sample representing the average of four cores covering anarea of 400 m2. Samples were taken in between 1995 and 1998.Trends show decreasing variability of soil organic carbon contentwith depth, but no significant changes in the variability of 114Ccontents with soil depth. Analytical errors are smaller than symbolsizes and are not visible (between 6–8 ‰ 114C).

Table 2. Overview variability of 114C and standard deviationacross regions and horizons converted to mean residence time(MRT) and coefficient of variation (CV).

LF (n= 12) 0–5 cm (n= 12) 5–10 cm (n= 10)

114C 159± 36.4 130± 28.6 67.4± 35.9Fm 1.17± 0.037 1.14± 0.029 1.08± 0.035CVFm 0.031 0.025 0.033MRT (yr) 7.58± 3.5 66.9± 23 143± 57CVMRT (yr) 0.46 0.34 0.40SOC (%) 37.93± 11.4 11.68± 7.26 6.40± 4.66CVSOC (%) 0.30 0.62 0.73

samples and a positive correlation between 114C and relief(slope) in samples from 0 to 5 cm depth. Furthermore, Spear-man correlation does indicate a significant positive correla-tion between 114C and SOC stocks in the 0–5 cm mineralsoil, and conversely a significant negative correlation in the10–20 cm depth interval. There was no significant correlationbetween 114C and Cation Exchange Capacity (CEC), but asignificant negative correlation between 114C and Fe con-tent in the 10–20 cm depth interval (Table 5). Mixed-effectmodels reveal few significant effects of environmental vari-ables on 114C at the 12 LWF sites except SOC (but not at5–10 cm) and MAT and Fe content at the 0–5 cm depth inter-val (Table 6).

-5

0

5

10

15

20

25

-300 -200 -100 0 100 200 300

0 10 20 30 40 50

Δ14C

Gley horizon (10-25 cm)

A horizon (5-17 cm)

Alptal NITREX site

organic layer

C:N ratio

C:N

-5

0

5

10

15

20

25

-300 -200 -100 0 100 200 300

0 10 20 30 40 50

Δ14C

0-5 cm

5-10 cm

Alptal LWF site

depth (cm)

10-20 cm

C:N ratio

C:NA B

Figure 4. Intra-forest variability (a) four samples each taken froman area of approximate 400 m2, totaling to 1600 m2 (b) four sam-ples representing micro-topographic end members sorted by hori-zon covering around to 1500 m2. The vertical bars represent thedepth intervals. The dashed line indicates the114C signature of theatmosphere at the time of sampling in resp; 1998 (a) and 2002 (b)(Hua et al., 2013).

3.2 Plot-scale (10-100 m) variability

In the subalpine spruce forest (Gleysol, Alptal catchment)two similarly sized plots were compared to assess inter-plotand intra-forest variability. Results show essentially similarranges and trends with soil depth for both sites, both in termsof 14C and C : N ratio. In the deeper soil horizons, the NI-TREX site has generally lower but also more variable 114Cvalues and slightly lower C : N ratios. Due to the differencesin horizon vs. depth sampling procedures further statisticalanalysis is however not possible (Fig. 4).

Plot-scale spatial variability was investigated at three LWFsites exhibiting the greatest (Podzol and Gleysol) and least(Cambisol) heterogeneity in terms of micro-relief (Fig. 5).Based on these end-members, we set brackets on the min-imum and maximum spatial variability in composition thatwe expect over the entire suite of study sites. These mea-surements reveal that on a scale of 10 m, standard deviationassociated with intra-site variability in 114C values can ex-ceed analytical error by tenfold. In the selected plots, thegreatest within-plot variability in 114C values among theeight 0.5 by 0.5 m soil pits, each taken in subquadrants of



0

10

20

30

40

50-100 0 100 200

Δ14C

Mixed samples 400 m2

Mixed sample 1600 m2

Statistical average

Cambisol

Single samples0

10

20

30

40

50-100 0 100 200

Δ14C

Mixed sample 1600 m2

Mixed samples 400 m2

Statistical average

Podzol

Single samples

SOC

(%)

0

10

20

30

40

50-100 0 100 200

Δ14C

Statistical averageMixed samples 400 m2

Gleysol

Single samples

r2 = 0.60

r2 = 0.00

r2 = 0.78

Figure 5. Relationship of 1 14C vs. soil organic carbon (SOC) content in topsoil samples (0–5 cm) from three sites with decreasing micro-topographic relief from left to right. Error bars indicate analytical error. The pattern shows that single samples can be unrepresentative butcareful physical homogenization of soil samples yields almost identical values than the arithmetic average of individual samples. The dashedvertical line in black indicates the 1 14C signature of the atmosphere at the time of sampling (Hua et al., 2013). The dotted lines representthe linear regression of 114C and SOC contents for single sample measurements.

10× 10 m (Fig. 2), ranges from 50 ‰ (relative highest de-gree of variability scenario, Podzol) to 20 ‰ (relative low-est degree of variability, Cambisol) (Table 3). In Supple-ment Fig. S1 the micro-topography is shown using digitalterrain mapping-derived curvature data. Slope is positivelycorrelated with the fraction modern coefficient of variation(CVFm) of single cores for the three high-resolution (Cam-bisol, Gleysol, Podzol) sites (r2

= 0.99 for n= 3). The SOCconcentration in the single cores of the Cambisol and Pod-zol correlate positively with114C signature (resp. r2

= 0.78,n= 7 and r2

= 0.60, n= 8), but for the Gleysol there is nocorrelation (r2

= 0.00) (Fig. 5). Pearson correlation betweenvariability (expressed as CVFm) and both environmental vari-ables (MAT, NPP, Elevation, Slope, SOC content, C : N ratio)yielded weak correlations (r<0.35), except for clay content(r = 0.71) and MAP (r = 0.51).

In order to assess the efficacy of mixing of samples toachieve representative samples, four composite samples fromthe quadrants were measured (each ∼ 400 m2) as well asa composite of 16 samples (encompassing the entire plot,∼ 1600 m2) as shown in Fig. 5. For the Cambisol and Pod-zol, the difference in 114C values between physically mixedbulk samples and the statistical average of individual sam-ples is close to or within analytical uncertainty (resp. 15.1and 0.46 ‰, Fig. 5) and considerably smaller than the stan-dard deviation of the quadrant samples. Semi-variograms didnot yield satisfactory results, as the variance did not plateauon the involved spatial scales.

Table 3. Intra-plot variability for three spatial variability end-members, expressed in 114C and standard deviation across regionsand horizons converted to mean residence time (MRT) and coeffi-cient of variation (CV). Variability is the highest for the Podzol andGleysol, and lowest for the Cambisol.

Podzol 0–5 cm Gleysol 0–5 cm Cambisol 0–5 cm(n= 7) (n= 8) (n= 8)

114C (‰) −1.48± 43.26 126± 35.2 129± 18.5Fm 1.00± 0.047 1.13± 0.038 1.35± 0.020CVFm 0.047 0.033 0.018MRT (yr) – 73.0± 27 2.64± 2.1CVMRT – 0.37 0.78

3.3 Influence of micro-topography and waterloggingon 14C variability

The assessment of within-plot variability related to micro-topography was investigated at the NITREX plot. Cores weregrouped according to morphological characteristics of theterrain:

1. Mounds (relative elevation up to 50 cm) with deeper wa-ter table, with a mor-type organic layer and an oxic Gleyhorizon.

2. Weak depressions (relative depression up to 15 cm) witha shallow water table, an anmoor-type organic layer anda partly (mottled) anoxic Gley soil.



-300-200-100 0 100 200 300

Δ14C

strong depression

intermediate typeweak depression

mound

Microtopography end-members

-10

0

10

20

30

0 10 20 30 40

SOC (%)

strong depression

mixed typeweak depression

mound

depth (cm)

12 14 16 18 20

C:N

C:N ratio mound

C:N ratio mixed typeC:N ratio strong depressionC:N ratio weak depression

-300-200-100 0 100 200 300

Δ14C

strong depression


mound


-10

0

10

20

30

0 10 20 30 40

SOC (%)

strong depression


mound

depth (cm)

12 14 16 18 20

C:N

C:N ratio mound


-300-200-100 0 100 200 300

Δ14C

strong depression


mound


-10

0

10

20

30

0 10 20 30 40

SOC (%)

strong depression


mound

depth (cm)

12 14 16 18 20

C:N

C:N ratio mound


-300-200-100 0 100 200 300

Δ14C

strong depression


mound


-10

0

10

20

30

0 10 20 30 40

SOC (%)

strong depression


mound

depth (cm)

12 14 16 18 20

C:N

C:N ratio mound


-300-200-100 0 100 200 300

Δ14C

strong depression


mound


-10

0

10

20

30

0 10 20 30 40

SOC (%)

strong depression


mound

depth (cm)

12 14 16 18 20

C:N

C:N ratio mound


Figure 6. Variability in SOC, 114C and C : N values between sep-arate micro-topography end-members (0–25 cm depth, in three in-tervals) of soils from the Alptal Valley. Weak depressions yield thelowest 114C values, whilst the mounds are most enriched in 14C.Analytical errors are smaller than symbol sizes and are not visible(between 6–8 ‰114C). The dotted vertical line for114C indicatesthe atmospheric input at the time of sampling (2002) (Hua et al.,2013).

3. Stronger depressions (relative depression up to 25 cm)with a very shallow water table, an anmoor-type or-ganic layer and underlying soil that is characterized byan anoxic Gley horizon.

4. Intermediate type (between type 1 and 2) with flat top,intermediate water table and anoxic Gley horizon. InFig. S2 the micro-topography is shown using digital el-evation mapping data.

Radiocarbon data combined with SOC contents and C : Nratios show (Fig. 6) that (1) contrasts exist between soils de-veloped under different microtopographic features with themound (oxic Gley) having the highest 114C value in thegley layers, indicating a higher degree of incorporation oforganic matter containing bomb-derived radiocarbon. Thestrong (anoxic Gley) and intermediate depression (partiallyanoxic Gley) exhibit very similar radiocarbon signal in theorganic layer, but with depth showing increasingly lower114C values on the mound, with the weak depression beingthe most 14C depleted in the Gley. 114C values for the in-termediate soil type falls between mounds and depressions.Additionally (2) SOC concentrations are higher in the top-soil of the mounds, but a reversed trend occurs at greaterdepth where the weak depression has higher SOC contentsin the Gley horizon. Lastly (3) C : N ratios show a similarreversal, as the weak depression has a markedly lower C : Nvalue than the other micro-topographic features in the top-soil, while having a higher value in the gleyic horizon.

0.1

1

10

100-300 -200 -100 0 100 200

Δ14C

Δ14C Gleysol 0–5 cm

Δ14C Podzol 0–5 cmΔ14C Podzol 5–10 cmΔ14C Podzol 10–20 cm

Δ14C Cambisol 0–5 cmΔ14C Cambisol 5–10 cmΔ14C Cambisol 10–20 cm

Δ14C Gleysol 5–10 cmΔ14C Gleysol 10–20 cm

Δ14C Regosol 0–5 cmΔ14C Regosol 5–10 cmΔ14C Regosol 10-20 cm

Δ14C Calcisol 0–5 cmΔ14C Calcisol 5–10 cmΔ14C Calcisol 10–20 cm

SOC

(%)

Figure 7. Spatial variability with depth (composites of each fourcores, covering 400 m2) for five different soil types (Regosol, Cam-bisol, Gleysol Podzol and Calcisol) covering a gradient in eleva-tion and average air temperature (MAT 9.2–1.3 ◦C, elevation 800–1900 m). The 114C signature of atmospheric input at the time ofsampling in indicated by grey shading (Hua et al., 2013). Variabil-ity in radiocarbon contents persists with depth and overlaps amongsoil types, while carbon contents differ among soil types and differ-ences decrease markedly with depth.

3.4 Spatial variability with depth

In order to establish 14C variability as a function of depthin topsoils, four composite soil samples (designed to assesswithin-plot spatial variability, each composed of four sub-cores from ∼ 400 m2, that encompass a total plot area of∼ 1600 m2, Fig. 2) were examined for five sites, each com-prising different soil types, at three depth intervals (0–5, 5–10 and 10–20 cm) (Fig. 7). These analyses revealed that spa-tial variability, expressed in SOC content and 114C, is moremuted but persists at depth. For the four 400 m2 compos-ites, the standard error of spatial variability is only 1–3 timesthe analytical error, with the exception of the highest vari-ability end-member (Podzol). The use of composite sampleshelps in reducing the potential for anomalous and unrepre-sentative values (Figs. 5, 7). For examination of deeper soils(> 20 cm), samples taken along a single profile per site weremeasured (Fig. 8). Results show that regardless of markeddifferences in soil type (and associated processes that influ-ence SOM) and large climatic gradients, the depth gradient in14C and overall amplitude of114C variations are remarkablysimilar. Two main exceptions are the Podzol, whose 114Cvalue in the topsoil is markedly lower than the other soils,and the erratic behaviour of the 114C trend in the Cambisolbelow 1 m depth. When taking into account the errors asso-ciated for single samples in the topsoil (Cambisol, Podzol,Gleysol), we see that the range of 114C values overlaps forfour soils (Cambisol, Podzol, Calcisol and Luvisol) within



-20

0

20

40

100200300400

0 1 2 3 4 5 20 40 60

SOC (%)

Depth (cm)

-900 -600 -300 0 300

Δ14C 0 10 20 30 40

C:N

Gleysol

Podzol

StagnosolCalcisol

Cambisol

Figure 8. Soil organic carbon (SOC) content, 114C and C : N ra-tios for five depth profiles of distinct soil types (Cambisol, Podzol,Calcisol, Stagnosol and Gleysol) covering a climatic and altitudinalgradient. Radiocarbon trends are highly similar for four out of fivesoil types, while carbon content and C : N ratio varies for each soiltype. The Podzol shows a different trend, with 114C consistentlylower in the topsoil. Analytical errors are smaller than symbol sizesand are not visible (between 6 and 8 ‰ 114C).

one standard error (Fig. 8). Regressing the 114C values withsoil depth to 100 cm depth showed tight linear relationships(r = 0.93–0.99). The depth gradients ranged between−3 and−8.5 ‰ cm−1 for the five sites, but showed no consistent re-lationship with any of the environmental variables (Table 1).The standard deviation associated with the SOC content andC : N ratios at these three locations also exceeds the valuesfor these parameters in the five profiles. Additionally, withthe exception of high-variability end-member Podzol, vari-ability of 14C values with depth exceeds lateral variability forthe intervals 0–20 cm. 14C variability with depth for deepersoils (> 20 cm) exceeds lateral variability for all soil types(Table 4).

4 Discussion

4.1 Uniformity in 14C signature across geological,climatic and depth gradients

Understanding the inherent stability and resilience of SOMin different climates is key for predicting how the carbon cy-cle may respond to future changes. The uniformity in topsoilradiocarbon signatures across the 12 sites spanning markedclimatic and geological gradients as well as soil types im-plies that the MRT of SOM in different ecosystems is rel-atively similar despite dissimilar environmental conditions.Few significant relationships emerged from comparison of114C values with environmental characteristics of the LWFsites (Tables 5 and 6). The significant positive correlation be-tween MAT and 114C for 0–5 cm depth found in the mixed-effect model (Table 6) indicates that the speed of incorpo-ration of bomb carbon is positively related to MAT, whichis consistent with the findings of Leifeld et al. (2015) in

Table4.V

ariabilityin

14Csignature

expressedin

Fmand

CV

forlateral(plot)andhorizontal(depth)variation

offivedifferentsoils.

PodzolG

leysolC

alcisolStagnosol

Cam

bisol

0–55–10

10–200–5

5–1010–20

0–510–20

0–55–10

10–200–5

5–1010–20

Fmlat

0.980±

0.0120.860

±0.017

0.816±

0.0351.12±

0.0191.08±

0.0180.909

±0.17

1.07±

0.0341.02±

0.0081.14±

0.0051.08±

0.0131.03±

0.0141.15±

0.0161.03±

0.0161.00±

0.007(n=

4)2

75

CV

Fmlat

0.0140.015

0.0070.017

0.0170.190

0.0320.008

0.0040.012

0.0130.014

0.0150.007

(n=

4)

TopsoilSubsoil

TopsoilSubsoil

TopsoilSubsoil

TopsoilSubsoil

TopsoilSubsoil

depthdepth

depthdepth

depthdepth

depthdepth

depthdepth

CV

Fmhor

0.988±

0.0320.0887

±0.18

1.08±

0.101.03±

0.14 b1.04±

0.0650.978

±0.10 c

1.10±

0.0491.02±

0.17 d1.06±

0.0481.05±

0.059

CV

Fmhor

0.0320.019

0.0970.13

0.0630.11

0.0440.17

0.0450.056

a0–65

cm, b

0–40cm

, c0–60

cm, d

0–70cm

.



grassland soils. However, the lack of correlation in the or-ganic layer and deeper layers indicate that confounding fac-tors such as variable ages from above- and belowground lit-ter input (Solly et al., 2013; Schrumpf et al., 2013) may im-pact the114C signal. The strong negative Spearman correla-tion of 114C and MAP at 10–20 cm depth implies a slowerturnover that could potentially be caused by increased wa-terlogging or anoxic conditions induced by higher precipita-tion. The significant positive correlation between 114C andSOC stocks may be indicative of soil layers that retain car-bon for longer (higher stocks) and therefore contain higherfractions of bomb-derived 14C entering the atmosphere fromthe 1950s onwards, as opposed to soils with lower stockswhich may have already lost the highly 14C labelled car-bon. The significant negative correlation of 114C and SOCstocks at 10–20 cm depth may reflect dilution of bomb-14Cderived carbon or delayed transmission of carbon containingthe bomb-14C signal in SOC-rich soils. Similarly, the nega-tive correlation between114C and iron content may indicatethat iron may contribute to carbon stabilisation and hencesmaller 114C value, which is consistent with recent findingsof Schrumpf et al. (2013) and Herold et al. (2014). The posi-tive relation between114C and relief (slope) for the shallow-est (0–5 cm depth) samples may be due to erosion-inducedhigher turnover rates that result in higher incorporation ofbomb-derived material. The lack of significant correlationbetween 114C and precipitation and primary production in-dicates that the incorporation of carbon in soils is not consis-tently dependent on these climatic factors. This is consistentwith findings of Doettler et al. (2015) that found that factorsother than climate exert strong controls on carbon stocks. Ithas to be noted that the former study concerned stocks only(i.e. not turnover), but in a steady state system it is reason-able to assume slower turnover is coupled to larger carbonstocks. Mathieu et al. (2015) also found that the carbon dy-namics in deeper soils are not controlled by climate but ratherby pedological properties, whereas topsoil carbon dynamicswere observed to be related to climate and cultivation. Fur-thermore, the high overall similarity in radiocarbon profileswith depth for five strongly contrasting soil types impliesthat this similarity persists with depth. The limited size ofthe data set prohibits more rigorous statistical analyses. Thisuniformity also indicates an insensitivity of 14C in SOM toenvironmental conditions. One explanation for this apparentinsensitivity could be that C input into soils and decomposi-tion are driven by similar factors; for instance increased litterinput with increasing MAT is accompanied by an increaseddecomposition rate. Furthermore, the presence of soil fauna(earthworms) at some sites with mull or modern type organiclayer (Bettlachstock, Schaenis, Lausanne, Alptal, Visp andNovaggio) may also complicate the response of carbon cy-cling to climate due to bioturbation (Ernst et al., 2008).

The 114C value for the Podzol in the topsoil is lower thanfor all other sites, which may be due to the unusually thick(> 20 cm) organic layer, which delays the transfer of the 14C

signal to the mineral soil. The erratic behaviour below 1 mdepth in the Cambisol (Lausanne) may be derived from amixture of syn-glacial and post-glacial carbon present in themoraine underlying the soil. The relatively small variabilityat the regional scale was even more striking as the corre-sponding CV was almost the same as at the plot scale for soilsamples located approximately 10 m apart from each other(Table 3).

There are only a few comparisons of 14C characteris-tics among soil types, but other studies have reported sim-ilar trends to those shown in Fig. 8 (Baisden and Parfitt,2007; Jenkinson et al., 2008). Scharpenseel and Becker-Heidelmann (1989) however, find very different 14C ages be-tween soil types for the depths 20, 50 and 100 cm on differ-ent continents. Mills et al. (2013) analysed literature data forforest top soils and found a similar range in values to oursamples. When comparing to cultivated soils we found largedifferences in 114C signature at depth (Paul et al., 1997).The variable time-point and horizon-notation of other soilradiocarbon samplings limit further comparison to existingliterature (Eusterhues et al., 2007; Paul et al., 1997; Rumpeland Kögel-Knabner, 2011; Scharpenseel et al., 1989). Thetime of soil formation may be paramount to the differencesthat develop in the bulk 14C age of organic carbon over time.Soil formation for the soils studied here initiated after thelast glacial retreat and can hence be assumed to have startedto form simultaneously, which may explain the similarity intheir 14C distribution with depth. Overall, the high similar-ity of 114C signatures between disparate soil types even atdepth indicates that the relative independence on tempera-ture and primary production may persist in deeper soils, andhence also for older SOM and during long-term C storage.

Further insights into the stability and vulnerability ofSOM will likely require investigations of specific fractions(Schrumpf et al., 2013) and compounds (Feng et al., 2008),where contrasts in the dynamics of SOM will likely be am-plified. Soil carbon turnover and stability models (e.g. CEN-TURY, Yasso, SoilR) that are highly dependent on inputparameters would also greatly benefit from improved esti-mates of spatial variability in 14C and other properties at theplot and regional scale, in combination with site-specific bio-chemical data (Liski et al., 2005; Schimel et al., 2001; Sierraet al., 2014).

4.2 Plot-scale variability and the influence ofmicro-topography

The marked plot-scale variability in 14C revealed in thisstudy is similar in magnitude to 14C signal variations ob-served across larger-scale gradients. In light of this obser-vation, developing and implementing sampling strategies toaccount for small-scale heterogeneity in SOM is essentialfor robust recognition and assessment of changes in SOMcomposition and stocks. It is also crucial that we better un-derstand factors that drive small-scale variability, as only



Table 5. Spearman correlation of 114C vs. climatic variables. Significant correlations are found between 114C against MAP and SOCat 10–20 cm. For p values p<0.05, p<0.005 and p<0.0005, values correlations are indicated with ∗, ∗∗ and ∗∗∗, p> 0.05 is indicatedwith superscript “ns”. Cation Exchange Capacity (CEC) in mmol kg−1 was calculated as defined by Blume et al. (2002). Iron (Fe) contentdetermined by HNO3 extraction in (mg kg−1) (Walthert et al., 2003).

MAT MAP NPP Elevation Slope SOC SOC stocks CEC Fe

Organic layer (n= 11) −0.31ns−0.34ns

−0.51ns 0.24ns−0.11ns

−0.47ns 0.76∗ −0.29ns 0.03ns

0–5 cm depth (n= 10) 0.29ns 0.10ns 0.08ns 0.04ns 0.65∗ −0.31∗ −0.42ns−0.41ns 0.47ns

5–10 cm depth (n= 11) −0.55ns−0.21ns 0.02ns 0.27ns

−0.2ns−0.29ns

−0.13ns 0.38ns−0.027ns

10–20 cm depth (n= 15) 0.12ns−0.78∗∗∗ 0.12ns

−0.12ns 0.31ns−0.59∗ −0.71∗ −0.37ns

−0.75∗∗

Table 6. Results of the mixed effect model testing114C for the effects of environmental variables. Site and core were always taken as randomeffects, whilst log(SOC), Elevation, MAT, MAP NPP, pH and log(Clay) are taken separately as fixed effects. The table shows F-values andsignificant effects are indicated are indicated with ∗, ∗∗ and ∗∗∗ for p values p<0.05, p<0.005 andp<0.0005. p > 0.05 are indicated withsuperscript “ns”.

Log(SOC) Elevation MAT MAP NPP pH H20 Log(Clay) CEC SOC stocks Fe

0–5 cm (n= 12 sites, n= 39 samples) 5.7∗ 3.8ns 4.9∗ 0.11ns 0.97ns 0.39ns 0.54ns 0.25ns 0.0055ns 7.9∗

5–10 cm (n= 4 sites, n= 15 samples) 0.23ns 2.3ns 1.1ns 0.27ns 1.5ns 0.34ns 1.3ns 1.6ns 0.51ns 4.0ns

10–20 cm (n= 5 sites, n= 20 samples) 14∗∗ 0.41ns 0.12ns 0.64ns 0.53ns 0.43ns 0.32ns 0.0060ns 54∗∗∗ 0.0031ns

then will we be able to incorporate these factors in larger-scale models of SOM turnover. We find a positive correla-tion between 114C and SOC content when examining the0–5 cm intervals for individual samples from the 10× 10 msub-quadrants (Fig. 5). When looking at the variability be-tween composite cores at 0–5 and 5–10 cm, this relationshipdisappears and re-appears only at 10–20 cm depth (Fig. 8).However, the relationship between 114C and SOC differedamong sites, indicating that other factors than C pool sizecontributed to the small-scale variability (Fig. 5). One reasoncould be spatially varying 14C inputs from plants which as aconsequence of differences plant functional types, plant/treeage, and between plant components such as roots, stems, nee-dles and leaves (e.g. Solly et al., 2013; Fröberg et al., 2012).The Cambisol, which exhibits a tight relation between 114Cand SOC, is located in a relative homogeneous beech for-est. In contrast, there is a poor correlation between 114Cand SOC in the Gleysol, and is characterized by a morecomplex vegetation mosaic with annual herbs in depressionsand perennial dwarf shrubs as well as spruce trees on themounds (Thimonier et al., 2011). 114C values of annualplants may differ substantially from perennial woody plants(Solly et al., 2013), and this could impart significant dif-ference in 14C input signals among microsites. The Podzolsite has an open canopy structure with dwarf shrubs domi-nating in the understory which may deliver a different 14Cinput signal than “aged” needle litter from trees. For thethree high-resolution sites (Podzol, Gleysol, Cambisol) weobserved that 14C variability was particularly high at siteswith steeper slopes, which are also marked by larger dif-ferences in micro-topography as well as higher occurrenceof soil disturbances induced by felled trees (windthrow). We

thus attribute the high variability in 114C values, especiallyfor the Podzol and Gleysol, to soil-mixing induced by tree-throw and subsequent higher erosion of loosened soil parti-cles. Micro-topographic contrasts can also be amplified bysoil erosion pathways (rills, ditches) affecting the hydrolog-ical properties such as the degree of waterlogging and waterrouting. Ernst et al. (2008) described the presence of earth-worms in the Gleysol and Cambisol, but not in the Podzol.Because of constraints on the data set size, no conclusivequantitative relationship can be established, but we hypoth-esise that the ubiquitous presence of in-soil fauna and asso-ciated transport activities would contribute to an overall in-crease in homogeneity rather than heterogeneity.

Geostatistical methods such as semi-variograms were ap-plied to the data but yielded unsatisfactory results becausethe variograms did not reach constant values. In terms ofmicro-topography, based on field observations, the measuredplots are representative for the larger areas. However, sam-pling and analysing a larger plot-size would be informativefor a semi-variogram analysis.

In the subalpine spruce forest (Gleysol, Alptal catchment),clear differences in radiocarbon signature can be seen be-tween different micro-topographic features (Fig. 6), confirm-ing prior observations of compositional variations (Schleppiet al., 1998). As would be expected, mounds that are con-ducive to less waterlogged (more oxygenated) conditionscontain more bomb 14C-containing (younger) carbon as rootand above-ground inputs are higher in these areas. Theanoxic depressions would be expected to receive the leastinput of biomass and preserve the material most optimallycompared to more oxygenated soil types. However, contraryto this expectation, the ‘intermediate’ weak depression with



a mottled oxygenated gley horizon, consistently holds thelowest 114C signature and exhibits the highest C : N ratioswithin the gley horizon despite similar SOC contents to theother microsites. A lower C : N ratio is widely regarded as aproxy for a higher degree of SOM degradation (Conen et al.,2008). Overall, the geochemical characteristics are indica-tive of decreased carbon degradation in the partially oxic sitecompared to the more fully oxic or completely anoxic sites.This may reflect different responses of microbial communi-ties to oscillating redox conditions or other properties asso-ciated with intermittent waterlogging. In this context, dif-ferences in hydrologic flow path (vertical, lateral) and resi-dence times (flow vs. stagnation) may be important. An addi-tional factor induced by micro-topography is the vegetation-induced differences in lag-time of the transfer of the atmo-spheric radiocarbon signal to SOM. The depressions supportgrowth of vascular plants with annual shoots and youngerfine roots, which may transfer the atmospheric signal fasterthan mound areas dominated by spruce trees and shrubs(Solly et al., 2013).

5 Conclusions

Variability of SOM signatures, notably radiocarbon, in shal-low and deep horizons of soils spanning large geologic andenvironmental gradients is similar to that observed at theplot-scale. Spatial variability could not be linked to clearenvironmental drivers (climatic and soil properties), exceptclay content and MAP. On a regional scale, there was also noapparent strong climatic driver for114C, except a weak rela-tionship with MAT for the shallowest (0–5 cm depth interval)soil horizons. While the present observations remain limitedin geographic scope, the relative homogeneity of 114C sig-natures observed in surface and deep soils across climaticand geologic gradients implies that the rate of C incorpora-tion may be similar and hence relatively insensitive to chang-ing climate conditions. The strong similarities in radiocarbontrends with depth among the different sites support the valid-ity of the extrapolation of currently sparse 14C observationsfrom the topsoil to deeper soil horizons in carbon modellingstudies. Systematic sampling of soils over areas that incorpo-rate the micro-topographic variability will reduce the impactof outliers and yield representative samples. The latter is es-sential for the use of radiocarbon to assess carbon turnoverand associated processes in forest soils, especially for plot-scale modelling. Future 14C investigations should includeexpansion of assessments of spatial variability to a broaderrange of soils, extension of such comparisons to deep soilhorizons, and in-depth investigations of variability amongSOM fractions and individual compounds.

The Supplement related to this article is available onlineat doi:10.5194/bg-13-3427-2016-supplement.

Author contributions. Tessa Sophia van der Voort selected the sam-ples, performed the measurements and large portion of the statis-tical analysis. Frank Hagedorn coordinated the statistical analysisand provided scientific feedback. Cameron McIntyre coordinatedthe 14C measurements and associated data processing. Claudia Zellaided in the measurements of samples for 14C. Lorenz Walthert andPatrick Schleppi collected samples or coordinated sample collec-tion. Xiaojuan Feng contributed to the project set-up and planning.Timothy Ian Eglinton provided the conceptual framework and aidedin the paper structure set-up. Tessa Sophia van der Voort preparedthe manuscript with help of all co-authors.

Acknowledgements. We would like to thank Susan Trumbore andtwo anonymous reviewers for their helpful and valuable comments.We thank the Swiss National Science Foundation (SNF) and theNational Research Plan 68 “Sustainable use of soils as a resource”(NRP68) for funding this project (SNF 406840_143023/11.1.13-31.12.15) in the framework of the SwissSOM “Sentinels”. Wethank the WSL LWF team for the sampling and data acquisition,and we would also like to thank the Laboratory of Ion Beam Physicsgroup of ETH Zurich, specifically Lukas Wacker for enabling themeasurements. Many thanks also to the other SwissSOM teammembers in this project: Sia Gosheva, Beatriz Gonzalez and CedricBader for their constructive input and synergy. Many thanks also toEmily Solly for her helpful comments and input. Last but not least,a special thanks to Flurin Sutter for his help with the assessment ofmicro-topography in the WSL LWF sites.

Edited by: Y. Kuzyakov

References

Baisden, W. T. and Parfitt, R. L.: Bomb 14C enrichment indicatesdecadal C pool in deep soil?, Biogeochemistry, 85, 59–68, 2007.

Batjes, N. H.: Total carbon and nitrogen in the soils of the world,Eur. J. Soil Sci., 47, 151–163, 1996.

Blume, H. P., Bruemmer, G. W., Horn, R., Kandeler, E., Kogel-Knabner, I., Kretzschmar, R., Stahr, K. and Wilke, B. M.:Lehrbuch der Bodenkunde, 16th Edn., 2002.

Conen, F., Zerva, A., Arrouays, D., Jolivet, C., Jarvis, P. G., Grace,J., and Mencuccini, M.: The carbon balance of forest soils: de-tectability of changes in soil carbon stocks in temperate and Bo-real forests, in: The Carbon Balance of Forest Biomes, 235–249,2004.

Conen, F., Zimmermann, M., Leifeld, J., Seth, B., and Alewell, C.:Relative stability of soil carbon revealed by shifts in δ15N andC : N ratio, Biogeosciences, 5, 123–128, doi:10.5194/bg-5-123-2008, 2008.

Davidson, E. A. and Janssens, I. A.: Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change, Nature,440, 165–73, 2006.

Doetterl, S., Stevens ,A., Six, J., Merckx, R., Van Oost, K.,Casanova Pinto, M., Casanova-Katny, A., Muñoz, C., Boudin,M., Zagal Venegas, E., and Boeckx, P.: Soil carbon storage con-trolled by interactions between geochemistry and climate, Nat.Geosci., 8, 1–4, 2015.


http://dx.doi.org/10.5194/bg-13-3427-2016-supplement

http://dx.doi.org/10.5194/bg-5-123-2008

http://dx.doi.org/10.5194/bg-5-123-2008


Ernst, G., Zimmermann, S., Christie, P. and Frey, B.: Mercury,cadmium and lead concentrations in different ecophysiologicalgroups of earthworms in forest soils, Environ. Pollut., 156, 1304–1313, 2008.

Etzold, S., Waldner, P., Thimonier, A., Schmitt, M., and Dobbertin,M.: Tree growth in Swiss forests between 1995 and 2010 in rela-tion to climate and stand conditions: Recent disturbances matter,Forest Ecol. Manag., 311, 41–55, 2014.

Eusterhues, K., Rumpel, C., and Kögel-Knabner, I.: Compositionand radiocarbon age of HF-resistant soil organic matter in a Pod-zol and a Cambisol, Org. Geochem., 38, 1356–1372, 2007.

Feng, X., Simpson, A. J., Wilson, K. P., Williams, D. D., and Simp-son, M. J.: Increased cuticular carbon sequestration and ligninoxidation in response to soil warming, Nat. Geosci., 1, 836–839,2008.

Fontaine, S., Barot, S., Barré, P., Bdioui, N., Mary, B., and Rumpel,C.: Stability of organic carbon in deep soil layers controlled byfresh carbon supply, Nature, 450, 277–280, 2007.

Fröberg, M.: Residence time of fine-root carbon using radiocarbonmeasurements of samples collected from a soil archive, J. Plant.Nutr. Soil Sc., 175, 46–48, 2012.

Goovaerts, P.: Geostatistical tools for characterizing the spatial vari-ability of microbiological and physico-chemical soil properties,Biol. Fertil. Soil., 27, 315–334, 1998.

Grüneberg, E., Schöning, I., Kalko, E. K. V., and Weisser, W. W.:Regional organic carbon stock variability: A comparison be-tween depth increments and soil horizons, Geoderma, 155, 426–433, 2010.

Herold, N., Schöning, I., Michalzik, B., Trumbore, S., andSchrumpf, M.: Controls on soil carbon storage and turnover inGerman landscapes, Biogeochemistry, 119, 435–451, 2014.

Hua, Q., Barbetti, M., and Rakowski, A. Z.: Atmospheric radio-carbon for the period 1950–2010, Radiocarbon, 55, 2059–2072,2013.

Innes, J. L.: Theoretical and practical criteria for the selection ofecosystem monitoring plots in swiss forests, Environ. Monit. As-sess., 36, 271–294, 1995.

Jenkinson, D. S., Poulton, P. R., and Bryant, C.: The turnover oforganic carbon in subsoils, Part 1. Natural and bomb radiocarbonin soil profiles from the Rothamsted long-term field experiments,Eur. J. Soil Sci., 59, 391–399, 2008.

Komada, T., Anderson, M. R., and Dorfmeier, C. L.: Carbonate re-moval from coastal sediments for the determination of organiccarbon and its isotopic signatures , δ13 C and δ14C?: compari-son of fumigation and direct acidification by hydrochloric acid,Limnol. Oceanogr. Methods, 6, 254–262, 2008.

Leifeld J., Meyer S., Budge K., Sebastia M., Zimmermann M., andFuhrer J.: Turnover of Grassland Roots in Mountain EcosystemsRevealed by Their Radiocarbon Signature: Role of Temperatureand Management, PLOS One, 10, 1–13, 2015

Levin, I., Naegler, T., Kromer, B., Diehl, M., Francey, R. J., Gomez-Pelaez, A. J., Steele, L. P., Wagenbach, D., Weller, R., and Wor-thy, D. E.: Observations and modelling of the global distributionand long-term trend of atmospheric 14CO2, Tellus B, 62, 26–46,2010.

Liski, J., Palosuo, T., Peltoniemi, M., and Sievänen, R.: Carbon anddecomposition model Yasso for forest soils, Ecol. Modell., 189,168–182, 2005.

Lutzow, M. V, Kogel-Knabner, I., Ekschmitt, K., Matzner, E.,Guggenberger, G., Marschner, B., and Flessa, H.: Stabilizationof organic matter in temperate soils: mechanisms and their rele-vance under different soil conditions – a review, Eur. J. Soil Sci.,57, 426–445, 2006.

Mathieu, J. A., Hatte, C., Balesdent, J., and Parent, E.: Deep soilcarbon dynamics are driven more by soil type than by climate: Aworldwide meta-analysis of radiocarbon profiles, Glob. Chang.Biol., 21, 4278–4292, 2015.

Melillo, J. M., Steudler, P. A., Aber, J. D., Newkirk, K., Lux, H.,Bowles, F. P., Catricala, C., Magill, A., Ahrens, T., and Morris-seau, S.: Soil warming and carbon-cycle feedbacks to the climatesystem, Science, 298, 2173–2176, 2002.

Mills, R. T. E., Tipping, E., Bryant, C. L., and Emmett, B. A.: Long-term organic carbon turnover rates in natural and semi-naturaltopsoils, Biogeochemistry, 118, 257–272, 2013.

Paul, E., Follett, R., Leavitt, W., Halvorson, A., Petersen, G., andLyon, D.: Radiocarbon Dating for Determination of Soil OrganicMatter Pool Sizes and Dynamics, Soil Sci. Soc. Am. J., 61, 1058–1067, 1997.

Pinheiro, J. C. and Bates, D. M.: Mixed-effect models in S and S-plus, Springer Science & Business Media, 2000.

Rumpel, C. and Kögel-Knabner, I.: Deep soil organic matter—a keybut poorly understood component of terrestrial C cycle, PlantSoil, 338, 143–158, 2011.

Scharpenseel, H., Becker-Heidelmann, P., Neue, H., and Tsutsuki,K.: Shifts in 14C patterns of soil profiles due to bomb carbon,inclusing effects of morphogenetic and turbation processes, Ra-diocarbon, 31, 627–636, 1989.

Schimel, D. S., House, J. I., Hibbard, K. a, Bousquet, P., Ciais, P.,Peylin, P., Braswell, B. H., Apps, M. J., Baker, D., Bondeau, A.,Canadell, J., Churkina, G., Cramer, W., Denning, a S., Field, C.B., Friedlingstein, P., Goodale, C., Heimann, M., Houghton, R.a, Melillo, J. M., Moore, B., Murdiyarso, D., Noble, I., Pacala, S.W., Prentice, I. C., Raupach, M. R., Rayner, P. J., Scholes, R. J.,Steffen, W. L., and Wirth, C.: Recent patterns and mechanisms ofcarbon exchange by terrestrial ecosystems, Nature, 414, 169–72,2001.

Schleppi, P., Muller, N., Feyen, H., Papritz, A., Bucher, J. B.,and Fluehler, H.: Nitrogen budgets of two small experimentalforested catchments at Alptal, Switzerland, Forest Ecol. Manag.,127, 177–185, 1998.

Schmidt, M. W. I., Torn M. S., Abiven S., Dittmar, T., Guggen-berger, G., Janssens., I. A., Kleber, K., Kögel-Kabner, I.,Lehmann, J., Manning, D. A. C., Nannipieri, P., Rasse, D. P.,Weiner, S., and Trumbore, S. E.: Persistence of soil organic mat-ter as an ecosystem property, Nature, 278, 49–56, 2011

Schrumpf, M., Kaiser, K., Guggenberger, G., Persson, T., Kögel-Knabner, I., and Schulze, E.-D.: Storage and stability of or-ganic carbon in soils as related to depth, occlusion within ag-gregates, and attachment to minerals, Biogeosciences, 10, 1675–1691, doi:10.5194/bg-10-1675-2013, 2013.

Sierra, C. A., Trumbore, S. E., Davidson, E. A., Frey, S. D., Savage,K. E., and Hopkins, F. M.: Predicting decadal trends and transientresponses of radiocarbon storage and fluxes in a temperate for-est soil, Biogeosciences, 9, 3013–3028, doi:10.5194/bg-9-3013-2012, 2012.


http://dx.doi.org/10.5194/bg-10-1675-2013

http://dx.doi.org/10.5194/bg-9-3013-2012

http://dx.doi.org/10.5194/bg-9-3013-2012


Sierra, C. A., Müller, M., and Trumbore, S. E.: Modeling radiocar-bon dynamics in soils: SoilR version 1.1, Geosci. Model Dev., 7,1919–1931, doi:10.5194/gmd-7-1919-2014, 2014.

Solly, E., Schöning, I., Boch, S., Müller, J., Socher, S. A., Trum-bore, S. E., and Schrumpf, M.: Mean age of carbon in fine rootsfrom temperate forests and grasslands with different manage-ment, Biogeosciences, 10, 4833–4843, doi:10.5194/bg-10-4833-2013, 2013.

Stuiver, M. and Polach, H. A.: Radiocarbon, Radiocarbon, 19, 355–363, 1977.

Thimonier, A., Kull, P., Keller, W., Moser, B., and Wohlgemuth, T.:Ground vegetation monitoring in Swiss forests: comparison ofsurvey methods and implications for trend assessments, Environ.Monit. Assess., 174, 47–63, 2011

Tipping, E., Billett, M. F., Bryant, C. L., Buckingham, S., andThacker, S. A.: Sources and ages of dissolved organic matterin peatland streams: evidence from chemistry mixture modellingand radiocarbon data, Biogeochemistry, 100, 121–137, 2010.

Tipping, E., Chamberlain, P. M., Fröberg, M., Hanson, P. J., andJardine, P. M.: Simulation of carbon cycling, including dissolvedorganic carbon transport, in forest soil locally enriched with 14C,Biogeochemistry, 108, 91–107, 2011.

Torn, M. S., Vitou, P. M., and Tru, E.: The Influence of NutrientAvailability on Soil Organic Matter Turnover estimated by Incu-bations and Radiocarbon Modelin, 8, 352–372, 2005.

Torn, M. S., Swanston, C. W., Castanha, C. and Trumbore, S. E.:Storage and turnover of organic matter in soil, in Biophysico-Chemical Processes Involving Natural Nonliving Organic Matterin Environmental Systems, John Wiley & Sons, Inc., p. 54, 2009.

Trumbore, S.: Radiocarbon and Soil Carbon Dynamics, Annu. Rev.Earth Planet. Sci., 37, 47–66, 2009.

Trumbore, S. E. and Czimczik, C. I.: Geology, An uncertain futurefor soil carbon, Science, 321, 1455–1456, 2008.

Von Lützow, M., Kögel-Knabner, I., Ludwig, B., Matzner, E.,Flessa, H., Ekschmitt, K., Guggenberger, G., Marschner, B., andKalbitz, K.: Stabilization mechanisms of organic matter in fourtemperate soils: Development and application of a conceptualmodel, J. Plant Nutr. Soil Sc., 171, 111–124, 2008.

Walthert, L., Lüscher, P., Luster, J., and Peter, B.: LangfristigeWaldökosystem-Forschung LWF in der Schweiz, KernprojektBodenmatrix. Aufnahmeanleitung zur ersten Erhebung 1994–1999, Swiss Federal Institute for Forest, Snow and LandscapeResearch WSL, Birmensdorf, 2002.

Walthert, L., Blaser, P., Lüscher, P., Luster, J., and Zimmer-mann, S.: Langfristige Waldökosystem-Forschung LWF in derSchweiz, Kernprojekt Bodenmatrix. Ergebnisse der ersten Erhe-bung 1994–1999, Swiss Federal Institute for Forest, Snow andLandscape Research WSL, Birmensdorf, 2003.

Wacker, L., Nemec, M., and Bourquin, J.: A revolutionary graphiti-sation system: Fully automated, compact and simple, Nucl. Instr.Meth. Phys. Res. Sect. B, 268, 931–934, 2009.

Wacker, L., Bonani, G., Friedrich, M., Hajdas, I., Kromer, B.,NÏmec, M., Ruff, M., Suter, M., Synal, H.-A., and Vockenhuber,C.: MICADAS: Routine and high-precision radiocarbon dating,Radiocarbon, 52, 252–262, 2010.

Walthert, L., Graf, U., Kammer, A., Luster, J., Pezzotta, D., Zim-mermann, S., and Hagedorn, F.: Determination of organic andinorganic carbon, δ13C, and nitrogen in soils containing carbon-ates after acid fumigation with HCl, J. Plant Nutr. Soil Sc., 173,207–216, 2010.

WRB: World reference base for soil resources 2006, World Soil Re-sources Reports 103, FAO, Rome, 2006.


http://dx.doi.org/10.5194/gmd-7-1919-2014

http://dx.doi.org/10.5194/bg-10-4833-2013

http://dx.doi.org/10.5194/bg-10-4833-2013

Documents

Variability in 14C contents of soil organic matter at …eprints.gla.ac.uk/136600/1/136600.pdf3428 T. S. van der Voort et al.: Variability in 14C contents of soil organic matter at